Skip to content

QUICK_START.md

Latest commit

 

History

History
175 lines (131 loc) · 3.92 KB

QUICK_START.md

File metadata and controls

175 lines (131 loc) · 3.92 KB

Quick Start Guide - MS/MS PLS-DA Analysis

5-Minute Start

Step 1: Install Packages (2 minutes)

# Install CRAN packages
install.packages(c("ggplot2", "ggrepel", "caret"))

# Install Bioconductor
if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("mixOmics")

Step 2: Prepare Your Data (1 minute)

You need two CSV files:

1. feature_matrix.csv - Your MS/MS intensities

              ,Sample1,Sample2,Sample3,Sample4,Sample5
mz_100.05     ,1234.5 ,1456.2 ,1389.7 ,1298.4 ,1367.9
mz_200.10     ,567.8  ,601.3  ,589.2  ,578.9  ,595.1
mz_300.15     ,8901.2 ,8756.4 ,8823.9 ,8912.7 ,8834.5

2. sample_annotation.csv - Your sample groups

SampleID,Condition
Sample1,Control
Sample2,Control
Sample3,Treatment
Sample4,Treatment
Sample5,Treatment

Step 3: Run Analysis (2 minutes)

Option A: Quick Start Script

source("MSMS_PLSDA_QuickStart.R")

Option B: Copy-Paste This:

library(ggplot2)
library(mixOmics)

# Load data
msms_data <- read.csv("feature_matrix.csv", row.names = 1)
annotation <- read.csv("sample_annotation.csv")

# Preprocess
msms_log <- log2(msms_data + 1)
msms_t <- t(msms_log)

# PLS-DA
plsda_result <- plsda(msms_t, annotation$Condition, ncomp = 2)

# Cross-validate
plsda_cv <- perf(plsda_result, validation = "loo", progressBar = FALSE)
accuracy <- (1 - plsda_cv$error.rate$overall[, 2]) * 100

# Get scores
plsda_scores <- as.data.frame(plsda_result$variates$X)
plsda_scores$Condition <- annotation$Condition

# Plot
ggplot(plsda_scores, aes(x = X1, y = X2, color = Condition)) +
  geom_point(size = 5) +
  stat_ellipse(level = 0.95) +
  labs(title = paste0("PLS-DA (Accuracy: ", round(accuracy, 1), "%)")) +
  theme_bw()

Done! You should see a PLS-DA plot with your data.

📊 What You Get

After running the script:

  • PLS-DA scores plot (saved as PNG)
  • Cross-validated accuracy
  • Top 20 discriminant features (VIP scores)
  • Sample predictions

Troubleshooting

Error: "cannot find function 'plsda'"

# Install mixOmics
BiocManager::install("mixOmics")
library(mixOmics)

Error: "object not found"

  • Check your file names match exactly
  • Make sure files are in working directory
  • Use getwd() to check current directory

Low accuracy (<70%)

  • Need more samples (try n≥20 total)
  • Groups may be too similar biologically
  • Try preprocessing data differently

All points in one cluster

  • Groups might not be separable
  • Check if you labeled samples correctly
  • Try running PCA first to check for patterns

Next Steps

  1. For comprehensive analysis: Use MSMS_PLSDA_Analysis.R
  2. Need to format data?: See Data_Format_Guide.R
  3. Want to understand more?: Read full README.md

Tips

  • Start with 2 components (ncomp = 2) - easiest to visualize
  • Use leave-one-out CV for small samples (n<30)
  • Check VIP scores > 1 for important features
  • Save your results before closing R
  • Document your parameters for reproducibility

Common Use Cases

Comparing two conditions (e.g., Disease vs Healthy)

annotation <- data.frame(
  SampleID = colnames(msms_data),
  Condition = c(rep("Healthy", 5), rep("Disease", 5))
)

Three or more groups

annotation <- data.frame(
  SampleID = colnames(msms_data),
  Condition = c(rep("Control", 4), rep("Treatment1", 3), rep("Treatment2", 3))
)

Very small sample (n=6, 3 per group)

# Use leave-one-out CV (automatic for small n)
plsda_cv <- perf(plsda_result, validation = "loo")

# Consider using only 1 component
plsda_result <- plsda(msms_t, annotation$Condition, ncomp = 1)

Need Help?

  1. Check Troubleshooting section
  2. Read FAQ
  3. Review mixOmics documentation
  4. Open an issue on GitHub

Good luck with your analysis! **lets make free science for everybody around the world.